ABSTRACT Survival for pediatric acute myeloid leukemia (pAML) has lagged behind other childhood leukemias due to the lack of effective pAML treatments. We have developed a novel T cell therapy, LV-10, made by transducing normal peripheral blood CD4+ T lymphocytes with a lentiviral vector expressing IL10. We discovered that LV-10 can lyse malignant myeloid cells via perforin and granzyme B. Of 23 primary pAML blast tested, 10 were lysed by all LV-10 lines tested, 6 were resistant , and 7 had a variable killing response depending on the LV-10 line to which they were exposed. Notably, the 6 patients who died had pAML blasts that were resistant or variably resistant to LV-10 lysis. The sensitivity of the pAML blasts to LV-10 lysis did not correlate with their genetic landscape, but RNA-sequencing (RNA-seq) of killing-sensitive and resistant blasts revealed dramatic differences in their transcriptomes. The molecular mechanism underlying the pAML resistance to lysis is yet unknown; resistance could be: 1. an active process impairing the LV-10 function, or 2. due to the absence of an antigen on pAML required for LV-10 recognition of target cells. In support of 1., RNA-seq revealed increased expression of OX40L and CD200 on resistant pAML blasts. Their respective receptors OX40 and CD200R1 are both expressed on LV-10, and may provide an inhibitory downstream signal. Thus, we hypothesize that resistant pAML blasts express protein(s) that mediate resistance to LV-10 lysis. However, the blasts with variable response to killing also suggest that the inherent heterogeneity of the different LV-10 lines may impact the efficiency of lysis. Thus, a second hypothesis states that lysis of variable blasts requires both LV-10 and pAML to have killing-permissive phenotypes. We will test these hypotheses by determining the lysis mechanism and pAML and LV-10 features required for LV-10 killing, paving the way for optimization of LV-10 for clinical use in pAML. To this end, we will test a new cohort of 45 pAML blasts for their sensitivity to killing, and their effect on LV-10 lytic capacity and activation. We will also ask if blocking OX40L and/or CD200 reverses the resistance, and if OX40 and CD200R1 signaling modifies LV-10 phenotype and function (Aim 1). In Aim 2, we will perform single cell RNA-seq on sensitive, resistant and variable pAML and LV-10, at baseline and after their interaction, to identify dynamic changes in both pAML and LV-10 population heterogeneity and gene expression, and uncover genes involved in resistance. More broadly, we note that the killing sensitivity of the tested pAML blasts did not correlate with conventional risk stratification methods such as WHO or FAB classification, cytogenetics, age, gender, leukocyte counts or minimal residual disease (MRD) +/- post-induction. This suggests that sensitivity to cytotoxic killing may offer a new means to stratify pAML patients. Importantly, the constitutive secretion of IL-10 implies allogeneic LV-10 can be tolerated in vivo, and may have wide application as a readily available, 3rd party off-the-shelf product administered as part of a hematopoietic stem cell transplantation, or as a stand-alone cell therapy forpAML.